| The crankshafts of automobile engine work in high temperature,high pressure and strong load.In order to meet the performance requirements,the high temperature forging in manufacturing process is usually adopted.Because of uneven distribution of the machining allowance and the wear of the mould for forging,the shape of the crankshaft may deviate from the design requirements.The existing methods of crankshaft inspection mainly adopts artificial scribing,which has low detection accuracy and low efficiency,and can not satisify automatic inspection.For this reason,a specific inspection software has been designed and developed,which is successfully applied to automatic optical inspection of crankshaft in industrial field,mainly including:(1)A specific software module of point cloud processing and contour parameter calculation for crankshaft detection is developed.Based on on-site requirements of workpiece inspection,geometry parameters of crankshaft detection is determined,and the calculation process of parameters extraction is designed;the artificial marking detection method is turned into software extraction method.The crankshaft measuring point detection software platform is developed and applied.The software modules include user interaction,pretreatment of point cloud,3D matching,extraction of 2D size,3D comparison and generation of report.(2)An iterative outlier removal algorithm based on neighborhood model is proposed.The neighborhood model of crankshaft measurement point is established,and the error function value of measurement point is calculated by design model matching.the isolated point cloud is identified and deleted according to the iteration termination condition;Through the experiment of many kinds of crankshaft measurement points,the range of the isolated points in the crankshaft measurement point is 0.2%-1.5%,and the distance error threshold is 3.8mm;Using the setting proportion and distance threshold the isolated point cloud of the crankshaft is iteratively calculated and recognized.the experiment shows that the proposed algorithm can identify the isolated point cloud in the crankshaft accurately,Moreover,the processing speed is improved.(3)A high efficient algorithm for large-scale crankshaft measurement is proposed.The crankshaft measuring point is segmented by OCTree method.then using the BSP tree to the point cloud segmentation again,select the cell division and cell surface variation minimum threshold as the judgment condition,features of the larger curvature section of the crankshaft,measuring overall lean rate is 30%,less time-consuming to streamline.Compared with the methods of random reduction and uniform mesh simplification,the time consumption of several simplification methods is little differente.The OCTree based on reduction method can retain the edge high curvature feature better.(4)automatic optical inspection of crankshaft is applied.The main process of software design for crankshaft measuring point software is designed,including preprocessing,point cloud matching,3D comparison,section 2D size calculation,test report generation and batch processing.Manual scribing test takes 30min-90 min,software automatic detection only takes10min-20 min.The subsequent work using batch operation instructions,automatic examination of a single piece of time only need 2min,greatly improved the efficiency of crankshaft detection. |
PDF Full Text Request